Nuclear Engineering and Design 67 (1981) 137-141 North-Holland Publishing Company 137 ANALYSIS OF PUMP START-UP TRANSIENTS R.B. G R O V E R a n d S.M. K O R A N N E Reactor Engineering Division, Bhabha Atomic Research Centre, Bombay 400 085, India Received 3 February 1981 A computer programme has been developed to study pump start-up transients. Predictions of this programme have been verified experimentaUy.Parametric studies indicate that an increase in fluid inertia increases the acceleration head, while an increase in the moment of inertia of rotating parts decreases the acceleration head. Also for systems where the ratio of moment of inertia of rotating parts to fluid inertia is low, pump speed goes marginally beyond the steady speed during the start-up transient. 1. Introduction During the starting of a centrifugal pump and prior to the time normal flow is reached, certain transient conditions can produce heads and consequently require torques much higher than design. In some cases selection of the driver and the pump must be based on starting rather than on normal flow conditions. If the system contains an appreciable amount of liquid, the inertia of the liquid mass could offer a significant resistance to any sudden change in velocity. Upon starting a primed pump and system without a valve, all the liquid in the system accelerates ,from rest to final condition of steady flow. The actual total system-head resistance curve at any instant during start-up will be the sum of the frictional resistance plus the inertial resistance. The inertial system head produced momentarily on starting, particularly high-specific speed pumps, is important when considering the duration of high driver torques and currents and the pressure rise in the system. To calculate the time to accelerate a centrifugal pump from rest and the total system resistance curve during start-up commonly used solutions neglect kinetic energy given to the fluid [1,2]. Boyd et al. [3] have analyzed transient flow performance in a multiloop nuclear reactor system taking into account the kinetic energy of the fluid, but they have mostly concentrated on flow coast down. They have given some results for pump start-up, but detailed parametric studies are not given. Also they have not compared their results with experiments. With a view to understand the influence on start-up of the two most important parameters, viz. inertia of the rotating parts and inertia of the fluid, a computer pro- gramme has been developed to predict the total system resistance curve from given design data. Predictions of this programme have been verified experimentally. This note gives the basic equations, the solution procedure, the experiment conducted and a few case studies. 2. Mathematical model When a pump is started, its driver has to provide torque for the following: (i) Overcoming frictional resistance offered to the flow of fluid by the piping, valves etc., (ii) Overcoming frictional torque for the rotating parts i.e. pump losses; (iii) Acceleration of rotating parts; (iv) Acceleration of fluid contained in the system piping. When the pump attains steady speed, no torque is required for acceleration. Torque is, then, needed only for overcoming pump losses and frictional resistance to flow of fluid in the system piping. Each of these torques is discussed hereafter: (a) Head required to overcome friction can be calculated by normal procedure. To overcome this head, driver has to supply torque Tf, which is given by Tf- pghrQ 20rN" (1) (b) Driver also has to provide torque to overcome pump losses. This data has to be obtained from the pump manufacturer and will take the following func- 0 0 2 9 - 5 4 9 3 / 8 1 / 0 0 0 0 - 0 0 0 0 / $ 0 2 . 7 5 © 1981 N o r t h - H o l l a n d 138 R.B. Grover, S.M. Koranne / Analysi,s of pump start-up transients tional form: Tpl = dp( N,Q ). (2) (c) Torque required to accelerate rotating parts is given by the following expression: TA,R = 2 ~ I ( d N / d t ) . (3) Substituting values of various torques from relations written earlier yields: Tm(U ) = 2 ~ r l - ~ + ~pghaQ + ~oghfQ -+qJ(N,Q). dN_ dt (d) Torque required to accelerate the fluid is given by the following expression: rA.swhere pghaQ 2,N ' ha (4) is the acceleration head and is given by h a _ I dQ g dt Iv" (5) I v is a measure of inertia of the fluid contained in the piping and is given by L~,j 1 1 I 2~r~ Tm(N) pgh,Q 2¢rU pghfQ } 2z'N q~( N , Q ) . (8) This equation has to be integrated to obtain the complete start-up transient. In order to integrate it, the i'elationship between Q and N must be known. Additional information is needed to determine this relationship. This relationship depends on driver speedtorque characteristics, pump characteristics and system characteristics. Driver speed-torque characteristics are normally available from the manufacturer or may be determined experimentally. Pump characteristics at the rated speed are available from the manufacturer and can be represented by the following equation: Hp(Q) = DQ z + BQ+ C. Using homologous theory, this may be generalized [4], ~ At, 1 i=1 j-I Here it is assumed that the piping system consists of n sections in series and the i th section consists of m parallel paths as shown in fig. 1. Pressure drop across the ith path d P i is related to flow, Q~./, through t h e j t h parallel path by the equation At steady state, Hp =h r , (10) while during start-up, d P, - ki.jQi, i . H p = h f +ho. During start-up, the following relation exists between various torques: Also at steady state, -- 2 T m = TA. R -}- TA, S q- Vf -4- rp. (7) dN/dt=O. (12) Eq. (8) therefore reduces to Tm(N~) - pghfQs 2~rN' "tl-2 71-1 Fig. 1. A piping network, showing system of numbering the branches. (11) ~-4,(Q~,N,). (13) Steady state flow rate, Q~, and steady state speed, N~, can thus be obtained from simultaneous solution of eqs. (9) and (13). All this information can then be used to integrate eq. (8). A computer programme called Pump Start-up Transient (PST) was written to numerically solve these equations. Predictions of the programme were compared with the experimental results obtained from a particular pumping system. This is discussed in the next section. R.B, Grover, S.M. Koranne / Analysis of pump start-up transients 139 3, Experimental verification ..... A simplified sketch of the moderator cooling circuit of a research reactor on which the experiment was conducted to verify the predictions of the computer programme is given in fig. 2. In this circuit pressure in the gas space above the moderator in the reactor vessel is held constant by a feed and bleed system. Level of the moderator in the vessel was kept constant during the experiments. This maintained the pressure at the bottom of the reactor vessel at a constant value during the start-up transient. Further, a pressure tap suitable for installation of a transducer was available only at some distance from the delivery of the pump and was used for recording the pressure transient. A strain gauge type transducer was used for converting the pressure signal into an electrical signal and was recorder on a fast recorder (make: Encardio-rite). For this circuit under steady state conditions, 30 12 /'2 = d P 2 f + P , and during start-up PE=dPEf +dP2a + P. The programme PST calculates head developed by the pump as a function of time. Pressure at any point in the circuit can be easily calculated. For example P2 will be given by P2 = k f d P t + k ~ d P a + P , // / 0 0.0 I I I OQ 1.6 24, I 3"Z TIME IN SECONDS Fig. 3. Comparison of predictions with experimental results. (14) and where kf- PREDICTIONSOF THE COI~ EXPERIMENTAL CURVE dP2f, s dpf, s , ka-- L2/.42 3 (15) (16) L,/A, i=1 GA$ LINES II V - ~ S T O R A G E TANK L ~ ,~,,,,~,,-ACONTROLVALVEAND <~a A GLOOEVAL~E 1 ~ : --" '--REACTORVESSEL Using these relations and the programme PST, pressure transient at the point 2 was calculated. Fig. 3 gives experimental results and also the predictions based on the code PST. Excellent agreement may be noted. An error analysis indicated that experimental results are within -+3.1%. I ~r-t- cONSTANTPRESSURE ] P 4. Case studies ] P2 /'~JI--BANK OF THREE IL~J~ PREHE:TREEXCHANGERS TRANSDUCER FOR THIS CIRCUIT I F s 131132~m"1 I = 0"2 k9. "m2 Fig. 2. Simplified hydraulic circuit. The computer programme was used to study the influence of moment of inertia of the rotating parts and fluid inertia on start-up transient. Fig. 4 gives head versus flow characteristics of a centrifugal pump used in the study. It also gives system head, acceleration head and total system resistance versus flow during start-up. 140 R.B. Grover, S.M. Koranne / Analysis of pump start-up transients NORMAL OPERATING POINT "' , / / I~TOTAL SYSTEM RESISTANCE / D U R I N G START-UP 1"50 o 1,35 I I----PUMP TOTAL HEAD AT RATED / 1 0-2 2 0-5 1"50 3 e,h < 1-35 IF 1-0 13832 an'~-1 - TOTAL SYSTEMRESISTANCE 1"20 .... ACCELRATION HEAD 1.05 - 0.90 ~0.gO i, Z o 0.75 zo <. I IN KS.nY" / / 1"20 1~ ? CURVE No. y /'q'-SYSTEM HEAD ('ALL F R I C T I O N } ~ o.~ 0"45 o ¢J ,tit. // /'~'~ ACCELRATION 0-75 1 0"60 "-% 0.45 \\x < 0.~ O'3C .... ~ ~ ~ \\ O 0'1~0 04 0.2 00 0"4 0.S 0"6 i 1.0 1.1 FLOW AS FRACTION OF RAT~'DFLOW Fig. 4. Transient system head during pump start-up for the system under study. 0"7 0'8 0.9 F r o m this figure it may be noted that total system resistance passes, through a m a x i m u m at 72.5% of rated flow. Acceleration head increases from zero to a maxim u m at 49.6% of rated flow a n d then again drops to zero. It was also n o t e d that speed goes beyond the rated speed by 0.214% at 0.9t seconds after start-up and then settles down to rated speed at 2.23 seconds after start-up. If the m o m e n t of inertia of the rotating parts is increased, the complete transient changes. Results for this case are given in fig. 5. If m o m e n t of inertia is very high, total system resistance no longer passes through a m a x i m u m , and p u m p speed rises to its rated value 0"00 I • . I I a a | I ~a 0'0 04 0"2 03 0.4 0,5 0.6 0"7 0'~ 0"9 1"0 FLOW AS FRACTIONOF RATED FLOW Fig. 5. Influence of moment of inertia of rotating parts on total system resistance and acceleration head during start-up. monotonically. Also the p u m p takes longer to reach the rated conditions. O n the other h a n d , if I F is increased, acceleration head increases and so m a x i m a in the total system resistance curve becomes more conspicuous; reverse occurs if I F is decreased as shown in fig. 6. Changes in I F do not have any appreciable influence on the rate of rise of speed, but it does influence the rate of rise of flow. Table 1 gives a summary of the results. It m a y be n o t e d that the a m o u n t by which speed overshoots, increases as I F is increased. However, speed c a n n o t go m u c h b e y o n d the rated speed because of the fact that in a n induction m o t o r m a x i m u m speed is limited by sync h r o n o u s speed. Table 1 Summary of results a Sr. no. Moment of inertia (kgm 2) IF (m- t ) Maximum speed ratio during transient Time after start-up when maximum speed occurs (sec) Maximum head developed during transient (sec) Time after start-up when maximum head is developed (sec) Total time to attain steady conditions (sec) 1 2 3 4 5 0.2 0.2 0.2 0.5 1.0 13832 27664 6916 13832 13832 1.00214 1.00578 1.00019 1.00006 1.00000 0.9105 0.8955 0.9655 2.482 5.327 61.93 64.86 58.59 58.16 57.80 0.8355 0.8505 0.8705 2.2250 4.7860 2.230 3.915 1.435 3.242 5.327 a Steady state h e a d : 57.79 m, steady state f l o w : 0.015729 m3/s. R.B. Grover, S.M. Koranne / Analysis of pump start-up transients -1 I F IN "m CURVE No. 1 6916 2 13632 3 2766/, 141 going through the manuscript and giving useful comments. Thanks are also due to the instrumentation groups of the Reactor Engineering Division and the Reactor Operations Division for their help in experimental work. MOMENT OF INERTIA'~ 0.2~.1m2 OF ROTATING PARTS~ 1.50 TOTAL SYSTEM RESISTANCE Q 1.35 ~ .... Nomenclature ACCELRATION HEAD 1-20 3 o.,o .o.Ts <[ .- o.,, 1" //'V: .o.,o I / . S - y 0-0C 0.0 /, __Z,. .... ...... \ ", ---.',, ' J ' ~ i J ~ ~ 0.1 0.2 0.3 0.4 0,5 0.6 0.7 0.8 FLOW AS FRACTION OF FLOW ! ! 0.9 1.0 Fig. 6. Influence of I v on total system resistance and acceleration head during start-up. 5. Concluding remarks The results presented in section 3 demonstrate the capability of the calculation procedure to predict pump start-up transients. From the case studies presented the following conclusions can be drawn: (a) For systems, where the ratio of moment of inertia of rotating parts to I F is low, not only pressure, but pump speed also goes marginally beyond the steady speed during the start-up transient. (b) Any change in I F does not appreciably influence the rate of rise of pump speed, while a change in moment of inertia of rotating parts does influence the rate of rise of pump speed. However, changes in both do influence the rate of rise of flow. (c) An increase in I v increases acceleration head, while an increase in the moment of inertia of the rotating parts decreases it. Acknowledgement The authors wish to thank Shri S.K. Mehta and Shri V. Venkat Raj of the Reactor Engineering Division, for A B,C,D dP g h I Iv k L N P Q t T to Area of cross-section of a pipe (m2) Constants in eq. (15) Pressure difference Acceleration due to gravity, 9.81 m / s 2 Head (m) Moment of inertia of rotating parts (kg.m2) A measure of fluid inertia defined by eq. (14). Constant defined by eq. (6). Length of pipe (m) Speed of rotation (rad/s) Pressure Flow rate (m3/s) Time (s) Torque (N.m) Density (kg/m 3) Subscripts 1,2,3, Sections of the circuit indicated in fig. 3 a A,R A,S f m P pl S Acceleration Acceleration of rotating parts Acceleration of system fluid Friction Motor Pump Pump losses Steady state References [1] I.J. Karassik et al., Pump Handbook (McGraw-Hill, New York, 1976) pp. 9-18. [2] L. Joseph and F.A. Hammil, Start-up pressures in short pump discharge lines, J. Hyd. Div., Proc. ASCE 98 (July 1972) 1125. [3] M.G. Boyd et al., Transient flow performance in a multiloop nuclear reactor system, Nucl. Sci. Engrg. 9 (1961) 442. [4] E.B. Wylie and V.L. Streeter, Fluid Transients (McGrawHill, New York, 1978) p. 104.